The FDA is having second thoughts about the broad labels it has granted PD-1 inhibitors in newly diagnosed stomach cancer, questioning whether restrictions should be placed on products from Bristol Myers Squibb and Merck & Co. plus a stomach cancer hopeful from BeiGene.
In a briefing document prepared for an Oncologic Drugs Advisory Committee meeting slated for Thursday, the FDA suggested that PD-1 inhibitors may not be suitable for certain patients with HER2-negative gastric cancer who have low PD-L1 expression, even though these immunotherapies have shown life-extension benefits in broad study populations.
After picking apart PD-L1 subgroup data from separate phase 3 trials of BMS’ Opdivo, Merck’s Keytruda and BeiGene’s Tevimbra, the FDA found that the drugs’ benefit-risk profiles “appeared not favorable” in PD-L1-negative patients and only “intermediate” in patients with tumors with PD-L1 combined positive scores or tumor area positivity scores below 10.
“Although these results are exploratory ... strong evidence does not appear to support the use of anti-PD-L1 drugs in patients with low PD-L1 expression,” the FDA said in its briefing document.
In 2021, Bristol’s Opdivo became the first PD-1 inhibitor approved, alongside chemo, as a front-line treatment for stomach cancer. Merck’s Keytruda followed suit in November 2023, and BeiGene’s application for the same indication for Tevimbra is currently under FDA review. Opdivo and Keytruda won broad FDA approvals in the disease covering patients regardless of their PD-L1 expression, and BeiGene is seeking the same for Tevimbra.
If the FDA gets its way, the agency’s stance would significantly reduce the number of stomach cancer patients that PD-1 drugs may target. Based on patient distribution patterns in its phase 3 trial, Merck noted that a selection cut-off at PD-L1 expression scores below 10 would exclude about 65% of new patients with HER2-negative stomach cancer, according to the company’s own briefing document.
Although the FDA’s existing approvals may be inclusive, the real-world use of PD-1 inhibitors in HER2-negative stomach cancer may not be as broad. The National Comprehensive Cancer Network guidelines give the highest category 1 recommendations to Opdivo only in tumors with PD-L1 expression scores of at least 5, and to Keytruda in cases with scores of 10 or above. Nevertheless, pointing to health records by Flatiron Health, Merck noted that about 25% of first-line stomach cancer patients were not being tested for PD-L1 expression before treatment.
The phase 3 trials the companies used to support their HER2-negative stomach cancer bids all showed statistically significant improvements in overall survival (OS) when all patients with various levels of PD-L1 expressions were counted. In the CheckMate-649 trial, adding Opdivo to chemotherapy cut the risk of death by 20%. In KEYNOTE-859, Keytruda’s addition to chemo led to a 22% reduction in the risk of death. The OS benefit was 20% for Tevimbra’s addition to chemo in the final analysis of RATIONALE-305.
However, the FDA found the OS benefit observed in the overall population “appears to be predominantly attributable to subgroups of patients with higher PD-L1 expression.”
The three trials enrolled patients with different characteristics. For its analysis, the FDA focused on HER2-negative gastric and gastroesophageal junction (GEJ) adenocarcinoma cases that are microsatellite stable or mismatch repair proficient.
Among patients with PD-L1-negative disease and PD-L1 scores below 10, the FDA analyses showed a death-risk reduction of 8% and 6%, respectively, for Opdivo; 8% and 14% for Keytruda; and 2% and 9% for Tevimbra.
The problems with the analyses were obvious, and the FDA didn’t shy away from their shortfalls: None of the trials were prospectively designed to measure the PD-1 inhibitors’ efficacy specifically in those PD-L1-negative or -low patients. Besides, the relatively small sample size for each trial’s subgroup also means their results may not be reliable.
“[H]owever, consistency of subgroup effects over multiple trials as well as biological plausibility can increase confidence in the subgroup results,” the FDA argued.
To add to the FDA’s doubts in the HER2-negative indication, Keytruda recently showed a potential detriment to OS in PD-L1-negative patients in the HER2-positive stomach cancer setting. A subgroup analysis of the phase 3 KEYNOTE-811 trial linked Keytruda’s combination with Herceptin and chemo to a higher risk of death versus Herceptin and chemo alone. Afterward, the FDA quickly limited Keytruda’s accelerated approval there to exclude PD-L1-negative patients.
Still, during a recent interview with Fierce Pharma, Marjorie Green, M.D., head of oncology global clinical development at Merck Research Laboratories, pointed out that the disease biology between HER2-negative and HER2-positive gastric cancer is different. She therefore argued that the HER2-positive finding should not be used to penalize Keytruda in the HER2-negative indication.
Besides, because the subgroups analysis for KEYNOTE-859 didn’t show any negative impact on OS, the results “suggest that there is a potential for benefit across all PD-L1 expression levels,” Merck argued in its own briefing document.
Overall, the FDA stated that the addition of PD-1 inhibitors to chemo “does not appear to result in benefit” in PD-L1-negative stomach cancer patients, while the benefit in patients with PD-L1 expression below 10 is unclear.
“If patients with low or no PD-L1 expression are not expected to benefit based on the available data, then administering anti-PD1 therapy has the potential for harm including serious immune related adverse events on top of a malignancy that can markedly affect a patient’s quality of life,” the agency said in its document.
The advisory committee will convene this Thursday, and the FDA’s external experts will weigh in on whether labeling of the drugs should be limited to a certain PD-L1 level. The agency is considering a classwide action because a drug-specific approach, while statistically sound, would cause “obstacles to the consistent treatment of patients with Ga/GEJ adenocarcinoma in the United States and in the conduct of future trials to improve outcomes of patients with gastric/GEJ adenocarcinoma.”
Besides stomach cancer, the FDA also raised a similar concern for PD-1 drugs in first-line esophageal squamous cell carcinoma based on findings from Merck’s KEYNOTE-590, BMS’ CheckMate-648 and BeiGene’s RATIONALE-306 trials, according to a separate briefing document for Thursday’s meeting.